read a large csv file and subset on datetime column

readcsv.py

from dateutil.parser import parse 

with open('data.csv') as fin:
    for line in csv.reader(fin):
       date1 = parse(row[1])
       date2 = datetime.now() - timedelta(days=3)
       if date1 > date2:
          continue
       process(row) # whatever your logic